Measurement of body fat with bioelectric impedance, skinfold thickness, and equations based on anthropometric measurements. Comparative analysis

BACKGROUND: There is growing clinical interest in estimating body fat percentages (%FM). The aim of this study was a comparison of body fat measurement by anthropometric equations, skinfold thickness (SFT) and bioelectrical impedance (BIA) methods. METHODS: Cross-sectional study. 149 healthy individuals (83 males and 66 females) were recruited. Height, weight, waist circumference, skinfold of biceps, triceps, subscapular and suprailiac regions, sum and log10 sum of four folds were obtained. %FM were calculated using the Siri equation (for sex or age-sex), Brozeck (sex or age-sex), Deurenberg, Lean and segmental BIA (Omron BF 300). Methodological differences among the various methods were analyzed with Spearman (SCC) and intraclass (ICC) correlation coefficients and Bland-Altman method. The reference method is the Siri-age-sex equation. RESULTS: The means of %FM were 26.8 +/- 8.3% (Siri-age-sex), 25.7 +/- 8.1% (Siri-sex), 25.97 +/- 7.6% (Brozeck-age-sex), 24.9 +/- 7.5% (Brozeck-sex), 25.6 +/- 8.8% (BIA), 28.6 +/- 8.2% (Deurenberg), 29.7 +/- 8.2% (Lean-waist circumference) and 29.4 +/- 9.3% (Lean-triceps-waist). The %FM values for all methods were highly intercorrelated (all SCC > 0.83 and CCI > 0.85). Brozeck-age-sex (average difference 0.8%; limits of agreement -0.48 to 2.08%) and BIA (1.17%; -6.21 to 8.55%) show the best agreement according to Bland and Altman analysis with Siri-age-sex equation. CONCLUSIONS: Deurenberg and Lean equations provide different body fat mass estimates than those derived from SFT measurement and BIA and thus should not be used interchangeably. The use of Siri-Brozeck equations is recommended for %FM assessment. The results suggest that segmental BIA-Omron BF 300 may be a valid alternative method.